The present invention relates to a monitoring apparatus for monitoring a video in an area to be monitored (monitoring subject area). In particular, the present invention relates to a parameter setting method by which a plurality of parameters used by an algorithm for detecting an object from a video can be adjusted easily by means of a GUI (Graphical User Interface).
It is conducted to monitor an object which intrudes into a monitoring subject area by using, for example, an image pickup device such as a television camera (TV camera). Furthermore, a technique for conducting monitoring which is not manned monitoring using a supervisor, but automatic monitoring using an apparatus or system.
As an example of a technique for automatically detecting an object which intrudes into a monitoring subject area, a monitoring technique using a method called background subtraction method has been widely used up to now. FIG. 4 shows an example of the background subtraction method. An object detection method using the background subtraction method includes the steps of calculating a difference in luminance (or pixel value) between an input image 401 obtained from an image pickup device or the like and a background image 402 serving as a reference in which an object to be detected is not displayed, binarizing the difference value in an obtained difference image 403 by using a predetermined threshold, and conducting monitoring by regarding that an object to be detected exists in an area which has a great difference and which appears on a binarized image 404, or there is a possibility thereof.
As methods for monitoring an object to be detected, there are an inter-frame difference method using differences among a plurality of input images obtained at different time points, and an optical flow method for detecting a motion flow in an image by calculating a motion of a local part between two images, besides the background subtraction method. A suitable method is implemented according to monitoring conditions for the monitoring subject area. In addition, movement of the detected area is tracked and a locus is obtained by, for example, continuously making a pattern search in the vicinity of the detected area. The kind of the object, such as a human being or an automobile, can be recognized on the basis of the velocity and size of the object which is being tracked.
In the monitoring apparatus, overlooking an object which has intruded and falsely recognizing a shaking motion of a tree as an object which has intruded often pose problems. However, it is possible to improve the precision of each image processing algorithm and alleviate the problems such as overlooking and false detection by adjusting parameters according to the input image. For example, a high processing precision is implemented by using the threshold in the background subtraction and the size of the pattern search range in the tracking as variable parameters and giving the parameters according to conditions of the peripheral environment. Furthermore, when the luminance difference between the object to be detected and the background is small due to foul weather such as a snowstorm, the precision of the detection can be raised by adjusting the threshold in the subtraction method to a low value.
As conventional techniques relating to the present invention, for example, there are techniques disclosed in JP-A-2005-062924 (hereafter referred to as Patent Document 1), US2008/0122949 (hereafter referred to as Patent Document 2), and JP-A-2006-114953 (hereafter referred to as Patent Document 3).